Vintage Synth Repair

>>I'm experienced enough to check the voltages, and come away with 3
brands, 4.37, 11.87 and 26.  Does that sound ok?  Is the 26 for
professional operation?<<

I wouldn't expect a 26V rail anywhere in a hybrid digital/analogue
circuit, but then they aren't all the same.  please bear in mind that
I'm writing this with no knowledge of this particular box or it's
schemos :-) I have a similar korg box here though, the sdd3300.

there may be a higher voltage for the display backlight but this would
probably be quite a bit higher (80V or so) & probably there'd be an
inverter to generate this close to the display itself.
again, I can't be certain on any of this, but it would be typical. for
all I know, the backlight could be a bunch of LEDs like in my old
cheetah sampler, or a fluorescent job with an inverter like in my emu
samplers.
 
so the 26V- you might be measuring across the whole of a +12 to -12,
of course.... no obvious negative voltages? 
& the 5 & 12 seem to be low, especially the 5. this might prevent the
CPU operating properly.

if you can read any of the chip numbers, especially things like audio
op-amps (5532, 4558, that sort of thing) then check their pin-outs on
a datasheet, you may discover more. 
the digital chips are where the problem probably is though, since the
box has scrambled brains, so you'd need to identify any of the logic
ICs & check those locally (i.e. instead of at the power supply); it
could be a cracked track on the back of the board, or a single bad
chip dragging a whole rail down. again, you might see familiar chip
numbers, 74xx or 474xx or even 4000-series stuff, 12 or 14 pins on
them, used as latches, gates & buffers. there are bound to be some of
these around the control panel.

thinking about that- you might want to dismantle the control panel &
check for contaminant ingress causing a shorted button- devices like
this can behave in a peculiar way sometimes if a button is "held down"
while they boot.

do you know what sort of power supply it is? is it a linear psu, with
a big mains transformer, bridges, big electrolytic caps & series
regulators? or a compact switched-mode device? the latter are easily
identified since they are so much smaller, & usually the host device
will work on any voltage from about 80 to 300, 50 or 60 Hz.
switched-mode psu's generally fail completely too.
if it's a linear power supply, it would usually have obvious output
regulators, either 78xx/79xx three-pin devices on heat sinks, or power
transistors performing the same function.

if you can find these devices & measure something "traditional" (12s &
5s, I mean) then the problem is probably with the CPU or the
associated ram. I had a juno 106 once that wouldn't start up because
the CPU ram was damaged; the memory battery was fine on that occasion,
but I had to replace the combined CPU & memory (with a revised design
comprising two separate devices that came separately by boat from
japan!)  

>>What do you think...beyond me?<<

I couldn't possibly say! :-) the aforementioned 106 was written off by
a professional synth service company & would be scrap but for my
persistence. do you know any decent techs near you? I don't know what
roland are like where you are, but they weren't much help to me; when
I wanted to fix the 106, I ordered a CPU & of course, the CPU turned
up 60 days later with no internal memory... 70 days after that I got
the thing working well enough to deal with it's many other issues.

oh well- hope this helps.

duncan.